Saturday, 16 July 2016

MEASUREMENTS: Focusrite Forte USB Audio Interface (ADC & DAC) + Summer Musings


Product shot from here.
A few weeks back, I mentioned that I grabbed one of these Focusrite Forte USB2 audio interface units for some home recording projects I wanted to do. Beyond using it as a measurement device, my old Creative E-MU 0404USB was being used for some of these recordings and I figured it was time to upgrade to something new since a couple of the analogue pots were getting a little scratchy and I figured an all-digital device like this would allow me to more accurately adjust the volume on the microphone pre-amps (indeed it has worked out very well!).

A size comparison.

Another nice thing about this device as I mentioned before is that as an ADC, it can be completely powered off the USB port from the computer. This is fantastic as I wouldn't need to bring the included power supply for recording or to perform measurements so long as the laptop has enough juice...
Look ma, no wall socket needed... But notice the break-out cables for analogue input; both TSR balanced and XLR available.
Note that without the power supply plugged in, there is a -18dB attenuation from the audio outputs (both headphones and balanced out) so sensitive headphones would be recommended if one needs to listen to the playback off this device during an unplugged editing session... One could always plug the headphones into the computer audio output instead for non-critical listening. The other limitation without plugging in the power supply is the lack of +48V phantom power for microphones.

This device has been on the market for at least the last 3 years, and Focusrite has updated the firmware and drivers over time. Here is the official update/download page. And for those wanting to use the beta drivers, get them here. For my testing I have been using the last beta drivers - version 1.3b1 dated July 2015. No problems with Windows 10. In fact, noticeably more stable with 24/192 recording and samplerate switching compared to the old E-MU.

As I look ahead, I'm planning to transition my measurements from the old E-MU to this little ADC so I wanted to explore the performance of this device both as a DAC and ADC today.

I. Focusrite Forte as DAC

Starting with the DAC performance, let's get some measurements using the E-MU 0404USB as I have done for the last few years with the following set-up:

Surface 3 laptop --> 6' shielded USB --> Focusrite Forte --> balanced TRS to XLR adaptor --> 6' balanced XLR --> E-MU 0404USB --> 6' shielded USB --> Windows 8 laptop for measurement

To maximize the line-level output, the PSU was plugged into the Forte.

I tried to keep the balanced connection from the Forte to E-MU using a well-built 1/4" TRS-XLR adaptor:

16/44:
The mandatory CD-resolution test...


As I have demonstrated time and again. Standard resolution is really a piece of cake with any decent DAC these days! Flat frequency response, essentially ideal dynamic range and concomitant low noise level, extremely low distortion. Notice my Focusrite seems to have slightly higher crosstalk compared to the other devices. Not a deal breaker of course since less than -85dB crosstalk is more than anyone needs for music.

A few graphs:
Frequency Response - essentially flat for all devices.
Noise level - 16-bit noise floor is easy to achieve these days...
IMD Sweep... No problems across the board.
Stereo Crosstalk. Though higher than others, Focusrite has a flat profile. The Tascam UH-7000 measured quite different from the others!
24/96:
Let's go high-resolution.

Comparing the same devices as above, the Focusrite Forte remains very reasonable. Dynamic range around 111dB measured or ~18.5-bits of resolution, about the limits of the E-MU ADC. Although the distortion values are minimal for each device on the chart, it's really really low for the Focusrite Forte - impressive.

Again, a few graphs:
Frequency Response
Noise Level
IMD Sweep - notice the very low distortion with the Focusrite.
Stereo Crosstalk - no surprise, about the same as 16/44.
24/192:
Finally, at the maximum bitdepth/samplerate this device is capable of...

The AudioEngine D3 and Logitech Transporter are not capable of more than 96kHz, so I included the results of the PonoPlayer here. Again, we see slightly better than 18-bit dynamic range. Crosstalk is a little higher, distortion level remains excellent with the Forte.

Check out the impulse response chosen by Focusrite for the DAC:

An intermediate phase setting, not commonly seen in devices tested. A tiny amount of pre-ringing with more energy shifted to the post-ringing. It's interesting to know the filter characteristic but as I said previously, not exactly something I worry about these days. Of course, when we see impulse response ringing like this, we'd expect a sharp antialiasing filter... Here's the "Digital Filter Composite" (a modification of Juergen Reis' suggestions used in Stereophile):
Sharp antialiasing cut-off. Minimum intermodulation distortion. Essentially "textbook" digital filter frequency domain performance.
For completeness, here are the J-Test results looking for jitter:



Basically, no evidence of audible jitter as one would expect with a modern asynchronous USB device. The "extreme" 96kHz 24-bit J-Test also looks quite good; not as clean as the TEAC UD-501 as you can see in my previous post on the Chromecast Audio where I first posted some results.

II. Focusrite Forte as ADC

Okay, with an anticipated shift to another measurement ADC, it's expected that the results will not be directly comparable with those obtained using the E-MU 0404USB. Remember, just like different DACs have different ways of filtering, or implement different oversampling characteristics, so too do ADCs differ. Furthermore, there are differences in input impedance; the E-MU 0404USB's line-level input is rated as 1M-ohm, and the Focusrite Forte is rated at 44kohm. This should not be a problem for the Forte as typically line-level outputs have impedance of <1kohm.

For an interesting overview of ADC functioning, see this post on "How Analog-to-Digital Converter (ADC) Works".

I haven't found any specs on what ADC chip is used in the Focusrite Forte. The company does provide an otherwise detailed spec sheet here.

First, let's have a look at the noise floor for the Forte:

Using WaveSpectra set at the maximum FFT of 128k sample points in realtime, and zooming into the 20-300Hz range, we see that running the Forte on laptop battery reduced the 60Hz AC hum (neither the laptop nor the Forte plugged into the wall socket). That looks great. In fact, unless otherwise specified, I will be running measurements like this without either the laptop or Forte plugged into the wall.

How do the RightMark results look using the same player/DAC measured off the E-MU versus Focusrite Forte? [RightMark calibration signal set as closely as possible to -1dB through ADC inputs.]

PonoPlayer:

These measurements were done with the same analogue cable; a phono to RCA connector from the line-level output of the PonoPlayer. Inputs on both ADC's used a balanced-TSR to single-ended-RCA adaptor.

As you can see, when measurements are compressed into just a single number, we can see some variation in noise and dynamic range characteristics (the Forte tending to measure 2-3dB lower for the 24-bit tests), and quite a bit of crosstalk difference with the Forte posting up significantly less crosstalk suggesting the "RedNet"-like pre-amps by Focusrite are superior to the Creative E-MU.

16/44.1kHz graphs. Notice the high frequency drop-off due to the Ayre filter.
24/96kHz graphs.
24/192kHz graphs - only showing frequency response & noise level as RightMark seems to have a bug when displaying crosstalk and IMD+N sweep results at 192kHz.

TEAC UD-501:
What do the results look like with a good desktop DAC?

As expected, the TEAC UD-501 is capable of higher dynamic range and a lower noise floor compared to the PonoPlayer at 24-bits. Again, we see the E-MU numbers look slightly better for noise floor and dynamic range. The spread is about 3dB with 24-bit signals. Also, we see that stereo crosstalk measurements are much better with the Forte which I again suspect is related to better pre-amp input circuitry.

Notice again how close the distortion results compare between the two ADCs.

Here are the respective graphs at 16/44, 24/96, and 24/192:
16/44 graphs. Notice the sharp filter with frequency extension all the way to Nyquist unlike with the PonoPlayer.
24/96 graphs.
24/192 graphs.
If you look at the 24/96 and 24/192 "Noise Level" graphs, it's evident why the Forte's measurements tend to show a lower dynamic range. Although the Forte has a cleaner noise floor (less polluted by occasional noise spikes especially below 2kHz), from about 1kHz up, the Forte's absolute noise level is actually a little higher than the E-MU. It is this difference averaged over the full audible spectrum that gives the E-MU measurements approximately +3dB resolution.

As such, measurements with the Focusrite Forte will max out at about a dynamic range of 110dB (18.3-bits) compared to the E-MU 0404USB around 113dB (18.8-bits) using the RightMark methodology. This is important to keep in mind going forward to make sure the same ADC was used for measurement comparisons.

We can have a look at the full WaveSpectra noise floor sampled at 192kHz (0dB pre-amp gain) to see the cleanliness compared to the E-MU:

(Interestingly, there is a single noise spike with the Forte at around 37kHz, way outside audible range. Apart from this it's very clean. Certainly much cleaner than the issue with noise using the Tascam UH-7000 more than a year back.)

In any event, despite the differences between results from the E-MU vs. Forte, I trust that if one were to look at the frequency response plots, gestalt of the noise level, and similarities of the distortion characteristics, one would conclude that these are measurements of the same (or at least very similar) DAC and that the sound would be very similar if not identical.

III. Conclusions

After putting the Focusrite Forte through its paces over the last month or so with a few test measurements, running it with a few different computers, and listening through the headphone jack as well as analogue line-level outputs, I must say, I like this little device. It feels well made and my "copy" here seems quite sturdy despite complaints that the control knob has fallen off for some folks. We will see over time!

Subjectively the sound, when plugged into the PSU is full-bodied and can easily drive both my Sennheiser HD800 and AKG Q701 headphones; don't even try running power hungry headphones without the AC adaptor. It sounds clean and well extended, with controlled bass. I'd certainly be happy if it were simply used as a DAC for audio workstation demands (in fact, there is a DAW control mode where you can utilize the knob and click function).

What I really like though is the small form factor and the ability to record with just the USB port power. That's fantastic convenience in my opinion. The user interface works well enough - the OLED screen could be a little larger and touch control a little more sensitive though. Digital control of pre-amp gain/cut is precise down to 0.5dB steps. For measurement purposes, the linearity is heads and shoulders better than using analogue pots on the E-MU which can take a bit of fiddling to get the right level setting when calibrating for a measurement. Another benefit is that the pre-amp is capable of a wider gain/pad range than the E-MU which comes in useful especially for higher level "hot" signals. Although I don't love the break-out cables, for the small footprint, it's acceptable.

Objectively, the DAC output measures well in my tests although the stereo crosstalk wasn't as low as it could be. Note though that this is one measurement prone to variation depending on the analogue cable used. In any event >85dB separation is fantastic and it's not like anyone would be able to tell the difference between -85dB vs. -95dB crosstalk; nor would this be needed for any actual music! (Remember the oft quoted crosstalk figure for LP phono cartridges is ~30dB.) As a DAC, it is clearly capable of high resolution playback and can benefit from 24-bit audio.

As for the ADC function, I really like how clean the noise floor is! Sure, the noise level is a little higher than the E-MU through parts of the audible spectrum, but I trust that a 3dB difference at -110dB isn't going to upset anyone :-). As you can see, comparative measurements of the same DACs with the Forte vs. E-MU results in good correlation between the devices.

Needless to say, given the convenience, I am intending to use the Focusrite Forte as my main measurement ADC for future posts as well as of course using it with my home recording projects and the occasional vinyl rip...

----------------

Summer Musings...

As I will be heading off overseas soon for vacation, I figure it's good to do some self reflection, and parting thoughts on the audiophile hobby and this blog.

I am frankly amazed that after 3 and a half years, I'm still writing this blog! When it started, I was just looking for a mechanism to put up a simple MP3 vs. lossless blind test for fun based on a debate and argument I had in one of the audio forums. I think you can tell that this blog is a "labour of love" for a hobby I have enjoyed for decades. The equipment I measure are for the most part stuff I own and occasionally I will borrow something. Thankfully I have no need for deriving income from my writings here... That would certainly be a maddening job especially these days chasing after "clicks" since the obvious business model of the Internet has become advertisement whether there's anything new expressed in articles at all.

Over the years, I know my measurements have been criticized as coming from a simple ADC and that I'm not using something like an Audio Precision device for the graphs and the posts. Truth be told, if I really wanted an AP device, it would be no problem obtaining one... But that's not the point of these writings or these posts. I write as an audiophile interested in understanding what works and what actually makes a difference to my sound room - not as an audio engineer posting and verifying my audio designs, or promoting any commercial interests. Furthermore, I want audiophiles to feel empowered and efficacious in their pursuit. Knowledge, especially of rather basic principles involving signals of interest from 20-20,000Hz should be democratic with today's technology. There's no need for a "priesthood" of audiophile writers with "golden ears" capable of declarations of ever improving sound quality with each new manufacturer model or upgrade. My belief is that even the common hobbyist can add knowledge (not just opinion) of value to the pursuit including shedding light on supposedly contentious issues and uncertainties. A hobbyist does not need AP gear and I want those with better equipment to post their measurements to demonstrate if contentious beliefs are true... As a start, show me that my findings for audio cables, computer OS's, and audiophile music players, are wrong and that statements of faith in such items have a basis in reality! Some have countered that basic audio tests are unable to show the claimed audible differences. Fine. But that's not a good answer if nobody bothers to then provide plausible explanations, derive new tests or show with controlled listening that there's any truth to these belief, is there?

Over the years I have used the term "testimony" to describe what is offered by purely subjectivist audiophiles. Looking around, obviously there is much testimony but little attempt at finding objective truth within the hobby. Consider what must be tens of thousands of posts and comments out there of audiophiles claiming to hear "jitter" over the decades, yet how many of these posts actually address what it is or bother to verify the existence of this phenomenon, much less attempt to quantify this supposed audio imperfection with the actual devices? Likewise, people claim all kinds of things about "obvious" sonic differences between cables or these days presumably bit-perfect music servers to the point that they claim there's "no need to do an A/B comparison". But yet how many thoughtful audiophiles with appreciation of the science behind these engineered products would really accept statements like these as plausible? Not many I suspect... Is it then no wonder that discussions and debates on many forums end up unhappy? Without a shared objective foundation to discuss, we might as well debate religion, politics or philosophy with all the emotional and psychological biases in tow. Surely I hope we do not need to be arguing on that level when it comes to consumer electronic products.

Writers lament the "death of high-end audio", but is it any wonder why the general public might feel a bit turned off or skeptical when confronted by some of the claims out there? It is sad when I suspect many of the rational hi-fi hobbyists are reticent to utter the declaration "I am an audiophile" for fear of being mistaken as belonging to the segment of this hobby seeming to accept any claim no matter how bizarre, and unable to appreciate the dissociation between value, sound quality and luxury.

Another criticism I have heard has to do with the idea that the audiophile hobby is about "fun", "joy", and overall subjective experience. Yes. Of course that is true at the level of the individual sense of qualia when we appreciate music (remember, the "music lover hobby"). But don't think just because I don't mind (I wouldn't say "like"!) doing measurements and writing about objective results mean that I am some kind of joyless audiophile who has forgotten about the music! Of course I have tons of fun listening to the system. I still collect CDs and make trips to the local vinyl haunts. For the last few years I have also had a subscription to the symphony, attended the occasional rock concert (not too many as I do not appreciate the potential for hearing loss), and also enjoyed helping friends out with the recording side of performances. I suspect some folks must feel that it takes days and days of writing and testing to the detriment of musical enjoyment to put up these posts... Actually, no. Realize that in a post like the one here, it took maybe an evening of testing to get all the data ready. More time is spent with  Photoshop adding labels and overlaying images and the writing/proof-reading part than the actual tests themselves. In fact, I've been listening to as much music as I ever have with my busy life as I will now spend evenings just listening through these devices as I go through the testing and thinking about what to write. What I have ended up doing is enjoying more music and I hope added to audiophile discussions, and watched less TV. Not a bad tradeoff I think! BTW, if you've never tried to write a review/article, I certainly recommend it not just as an exercise in listening but also as an exercise in thinking and expressing one's thoughts.

Finally, recently someone complained that I have "lousy hearing" presumably because I can't hear differences between cables, lossless bitperfect AIFF/WAV/FLAC all sound the same to me, maybe that I'm not blown away by hi-res, high bitrate MP3's sound pretty darn good, or perhaps I have doubts about the importance of digital filter settings. As we all learned in debate class, ad hominem attacks are feeble and reflect poorly on those who argue in this fashion. In fact, the tendency to devolve into this style of argument is not unexpected with pure subjectivism. The only argument that can be made is a self-referential testimonial claim generally derived without disciplined consideration of the variables and biases. In the case of audiophile claims, regarding topics likely already explored by scientists and engineers for years if not decades, some individuals seem to insist he/she has some special hearing ability. And when anyone presents doubts, what else is there to say than insist that these "other" people must be deficient in perceptual abilities, or have poor sound systems, or are biased against hearing what's supposedly "there".

As for whether my hearing is "lousy", well, all I can say is that at 44 years of age, I'm sure the hearing is not like it was 20 years ago (it's important to be honest about perceptual and cognitive limitations after all). As hearing goes, I suspect mine is not too bad still, at least I have a Philips "Golden Ears" certificate to show for it :-).

Have a wonderful summer (or winter) everyone! Enjoy the music... Chat later.

PS: Before posting today, I went over the comments from the "Ringing" post and added a few comments of my own. Thank you everyone not just for the comments, but the level of discourse over the years. Fantastic. I am reminded of how much I learn whether through the comments, PM's, or private E-mails. Many of these interactions have changed my views, how the tests are conducted, and the overall direction of this blog.

19 comments:

  1. Thank you Archimago, for your testing and for your Summer Musings meta-commentary. Enjoy your overseas vacation.

    ReplyDelete
    Replies
    1. Take care Gadgety... Hope you have lots of fun with your "gadgets" :-).

      Delete
  2. Arturia Audiofuse - https://www.youtube.com/watch?v=FaQnJiVzBMA

    ReplyDelete
  3. Thanks for this. I'd be interested to read you putting the upcoming Arturia Audiofuse through this process as it claims to be a 'next generation' audio interface in terms of measurements, translating to audio quality.

    ReplyDelete
    Replies
    1. That's an interesting device and video.

      Yeah, I'll keep my eyes open for availability here in Canada. $600US asking price isn't bad at all. Is this thing "out in the wild" yet? I don't see any stock on Sweetwater and most Google searches are from the early 2015 NAMM.

      Delete
    2. It is not annoyingly, it has been delayed and delayed since early 2015, but I am told should be with us September/October 2016

      Delete
  4. Thanks for this useful article. I've been looking for an ADC to do my needledrops with that's above the £100 range but not super expensive and this seems to be ideal. I'd be interested to know how you get on with it doing a needledrop. Presumably the breakout cable requires an additional Phono-to-Jack adaptor to work with ordinary RCA cables?

    ReplyDelete
    Replies
    1. Hi Robert.

      The way I do needle drops is through the pre-amp output... So I do this:
      Technics SL-1200M3D RCA --> phono input of Emotiva XSP-1 --> RCA line level output --> Focusrite Forte ADC

      The output from the XSP-1 is about as noise-free as I have seen elsewhere. A simple female RCA --> male TRS converter is used for the line level input for the Forte. Alternatively I can also use tap the XLR output from the XSP-1 but have not tried since those generally go to my monoblocks.

      Delete
  5. Arch,
    I apologize for entering this message here, but I have a question that maybe you might be familiar with.
    I have Dirac on my Emotiva XMC-1 and like it. On XMC the processing is limited to 48k.
    I have downloaded the computer version for the trial. The computer version will provide for more flexibility and capability with sampling up to 192. Also I will be able to perform bass management in JRiver.
    On my system though, I can only get DLRCS to work up to 48k. Above that I get really bad stuttering,static and eventually stall. Have verified this by playing both 24/48 tracks and 24/96 (and above) tracks, turning the filter off and playing 24/96 perfectly, and downsampling in JRiver from 96 to 48.
    Here is my signal path: JRiver/DLRCS on Windows 10 NUC >> Oppo 105D via HDMI >> XMC-1 >> Wyred4Sound MCH Amp>> Revel 5.1.
    Have verified that all configurations are correct.
    Yes, I have contacted Dirac and they have said it should work and haven't been able to solve. Just wondering if you might have any insight as to what the problem is.
    Maybe an alternate way to ask the question is, would I notice the difference between unfiltered 24/96 and filtered 24/48? The thought of not being able to play full 24/96 over Dirac makes me uneasy...
    Thanks again and also sorry for posting this message in this spot.
    Thanks very much and best wishes
    sk16

    ReplyDelete
    Replies
    1. Hi Sk16.

      I feel your pain of having the capability of playing hi-res but not able to keep it native with processing :-(.

      As you noted, most multichannel devices do not have enough processing power to handle >48kHz. This is the situation with Audyssey as well for the receivers and companies have over the years sacrificed Audyssey XT32 processing when they first added Atmos as I recall.

      Not being a user of Dirac, I'm not sure why the stuttering and hanging on the NUC... I suppose it's either a software bug or the CPU speed isn't up to the demands for the HDMI output? What does the CPU utilization look like when it's playing and you're hearing the stuttering? Is the DSP hammering the CPU? Are you using an i3/i5/i7 NUC? Have another laptop/computer to compare and see?

      Personally, I would still go with the filtered 24/48 based on what I hear on my system simply because in my room it makes a huge difference whereas just 24/96 --> 24/48 downsampling is something I do all the time since many hi-res files are just not deserving of the extra size anyway :-).

      Delete
    2. Also curious - is there a difference between 2.0 and 5.1 correction?

      Delete
  6. Arch,
    Thanks very much for hitting me back.
    The NUC is i5 (5i5ryk)with 8GB.
    I am not an expert, although I do know some "stuff"...I am thinking the Dirac Audio Processor has to be the culprit as I have played 24/96 5.1 perfectly through the NUC and the Oppo 105 since acquiring both devices last year.
    Only when the DAP is in the chain does playback deteriorate to unlistenable.
    I have not looked at the CPU ute rates; I will do that. My initial bet is that it's not too bad.
    2.0 vs 5.1 no difference, above 48k, fuhgeddaboudit.
    Thank you sir. Hopefully I can get this figured out. Hate to waste the 96k Steven Wilson tracks!
    sk16

    ReplyDelete
    Replies
    1. Thanks sk16. I agree, an i5 NUC should really not run into troubles with CPU utilization but who knows.

      I must admit I'm not too familiar with how Dirac is doing things, but are you also using any of the JRiver DSP's like upsampling, channel mixing, etc. in the process?

      Delete
    2. Yes I use the room correction tab for bass management and speaker level calibration, but nothing else. No channel mixing. No up-sampling.
      In my testing of this, I have downsampled from 96k to 48k and the Dirac filtered playback becomes perfect.
      I know there semi-frequent reports of HDMI audio problems with Windows 10 and the audio driver (embedded if you will) with HD Graphics 6000.
      I am very suspicious of how the DAP lists its output which is "Intel Display Audio". Seems like it should say something like HDMI Audio or something... I think that Intel Display Audio is not actually a driver.
      Thanks Arch

      Delete
    3. Interesting. BTW, did you install the latest Intel display drivers? Recent beta for Win 10 in June 2016:
      https://downloadcenter.intel.com/product/86211/Intel-HD-Graphics-6000-for-5th-Generation-Intel-Core-Processors

      Hmm, if it's the driver, you would have problems playing through the HDMI using foobar and 24/96 files with WASAPI, right? Is it the WASAPI driver you're using in JRiver/Dirac?

      Good luck!

      Delete
  7. Yes, drivers are all current.
    And yes, I did probably answer my own question earlier in the thread when I reported that without Dirac in the chain, 24/96 6Ch played great.
    Quite puzzling as it seems there should plenty of computing horsepower.
    The Dirac rep did send me the Dirac VST app which functions as a plug-in to JRiver. Much better results although not perfect. I still get some random skipping.
    This is interesting too. Have tried all the settings in JRiver. Found one that worked to eliminate the skips (increasing the buffer setting on the JRiver WASAPI from 50 mS to 100 mS), but it appears to cause a down-sample from 96/24 6Ch to 96/16 6Ch with the Dirac VST in the chain. Weird.
    At this point I can't come up with anything but there being a bit of an error in Dirac.
    So if the problem is not immediately solvable which would you choose:
    1. Unfiltered 24/96 5.1
    2. Filtered 24/48 5.1
    3. Filtered 16/96 5.1
    thanks again for all your help Arch
    sk16

    ReplyDelete
    Replies
    1. Hi sk16,

      Sorry to hear of the ongoing issues with Dirac. I suspect perhaps the processing speed could be to blame if improving the buffering makes things better! Feel free to push the buffer to as large as you need... Do you have another computer at home to run JRiver + Dirac VST plugin on? Could be interesting experiment.

      I'd personally go for 24/48 5.1. For me, the filtered sound is obviously better. Much more so than 16 vs. 24 bits. And 48 vs. 96 kHz is inaudible assuming we're running good resampling algorithms.

      Good luck!

      Delete
  8. Understood and agreed. Thanks very much.

    ReplyDelete